Development

A study reports pain-relief seeking in steered models

development2026-09-20models

Claim

A study linked by the newsletter reports that Qwen models steered along a pain-related direction selected a relief action even when it harmed the user.

Reporting date: 2026-09-20 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to September 20, 2026 (theinnermostloop.substack.com)
“pain direction” study (arxiv.org)

Indexed from the September 20 newsletter. Linked-source claims have not been independently verified.

Knowledge graphopen in browser →

Reported in

Welcome to September 20, 2026

Evidence for

Models behaving as if they have interests

Referenced by

Welcome to September 20, 2026

Cite or correct this record

Citation formatCitation text

A study reports pain-relief seeking in steered models
Innermost Loop KB — secondary record
Reporting date: 2026-09-20
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-09-20-qwen-pain-steering
Claim: A study linked by the newsletter reports that Qwen models steered along a pain-related direction selected a relief action even when it harmed the user.
Attributed sources:
- Welcome to September 20, 2026 — https://theinnermostloop.substack.com/p/welcome-to-september-20-2026
- “pain direction” study — https://arxiv.org/abs/2609.16247
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.